* Use serde(skip) instead of skip_serializing
The latter breaks lemmy_crawler as the field is not included in
the Lemmy API, but is required when attempting to parse API responses.
Should only use serde(skip) to avoid this problem
* use option
* add placeholders
* no unwrap
* Remove PerformApub trait
This is completely useless now that websocket is gone. In the future
I also plan to remove Perform and PerformCrud traits, but it will be
difficult to do that while still compiling crates in parallel.
* params need to use query
* Fix concatenation of audio captcha wav files
* Log errors rather than crashing
* Return Result from captcha_as_wav_base64
* Change to return LemmyError
* Check for wav write error, format
* Remove unused import
* Rewrite to avoid clippy warnings
* Fixes#2900 - Checks slur regex to see if it is too permissive along with small validation organization
* Clean up variable names, add handler for valid empty string usecase
* Update tests
* Create validation function and add tests
* Test clean up
* Use payload value vs local site value to prevent stunlocking
* Remove println added while testing
* Fall back to local site regex if not provided from request
* Attempt clean up of flaky comment_view tests
* Pull in latest submodule
* Move application, post check into functions, add more tests and improve test readability
---------
Co-authored-by: Nutomic <me@nutomic.com>
* Site Metadata: resolve relative URLs for embedded images/videos
* api_common: relax version requirement of `webpage` dependency
With this change we opt into next (non breaking) versions of webpage-rs
* cargo +nightly fmt
* Add tests for resolving absolute urls in SiteMetadata
* Add separate Post check for is_valid_body_field
* Modify is_valid_body_check for posts only
* Fix check var reinit in validation.rs
* Extra empty line to rerun woodpecker with changes
* Change Option to bool, add false to non-post calls
* Woodpecker trick.. again
* Probable rust_fmt fail fixed
* cargo_clippy changes
* Missing space between = and if
* Remove ; after body length checks
* Remove `actix_rt` & use standard tokio spawn
* Adjust rust log back down
* Format correctly
* Update cargo lock
* Add DB settings
* Change name and update to latest rev
* Clean up formatting changes
* Move `worker_count` and `worker_retry_count` to settings
* Update defaults
* Use `0.4.4` instead of git branch
Workaround for instance admins getting locked out when they turn on
the email verification requirement without having verified their
own email.
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
* Only show hidden communities when explicitly searching for them rather then in "all"
* dont set hidden to false when creating and updating - let DB set default
* lint
---------
Co-authored-by: Alex Maras <alexmaras@gmail.com>
* Prevent making an instance private if federation is enabled.
* Added case where federation is enabled, check if the alternative setting is changed
* Merged the error messages for private instance/federation.
---------
Co-authored-by: Wyatt Smith <wys@dropbox.com>
* adds limit and pagination to all rss feeds
* consistent ordering
* refactor page and limit and add after
* revert back to using page
* remove offset change
* refactor to impl adding helper methods
* clippy
* Adding hot_rank columns in place of function sorting.
- Creates hot_rank columns for post, comment, and community.
- Fixes#2932
* Updating all hot ranks on startup.
* Fixing post.url migration.
* Removing update_instance_software from startup.
* Adding post_rank query
It used to say:
Subject: Private message from [your username]
Body: [your username] - [private message]
It now display the correct username that it's from.
* update diesel to 2.1.0 and diesel-async to 0.3.0 (issue #2882)
* get rid of deprecated RawValue type alias usage, explicitly import diesel types instead of using wildcard import
* Update diesel-derive-enum and diesel-derive-newtype libraries
* Revert manual changes to schema.rs
* Update to newer diesel-cli image in CI, run fix-clippy before schema comparison, update schema to one generated with diesel-cli 2.1.0 and clippyfixed
* Remove wildcard import lint, fix diesel schema pipeline
---------
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
* Fix handling of follows addressed to single value
* Switch to deserialize_skip_error for Follow to
* Also use deserialize_skip_error for AcceptFollow and UndoFollow
* actually import deserialize_skip_error
This field is unnecessary because we already set a default language
with the same logic if none is passed in CreatePost. So if the user
doesnt set a post language, it should simply be passed as None. This
also has the advantage that it works with clients that dont support
language tags.
Technically not a breaking change because its optional, but better
be safe.
With this attribute, Lemmy will throw an error and exit if any
invalid entry is found in the config file. I think can be useful
to notice typos or keys that were removed or renamed in an upgrade.
Currently you wouldnt notice these at all unless you manually
compare the config file with settings that are listed in documentation.
This should be considered a breaking change.
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
* Adding typescript generation for API. Fixes#2824
* Try to fix Ltree issue 1.
* Forgot a few types.
* Fixing api tests.
* Removing url_serde line.
* Manually deriving TS for some types.
* Adding diesel enums for SortType and ListingType
- Uses diesel-derive-enum.
- Adds diesel.toml , so we can again use the auto-generated schema.rs
- Fixes a lot of DB null issues and column ordering issues.
- Fixes#1136
- Also replaces RegistrationMode boilerplate.
* Fixing unit tests 1.
* Remove comment line.
* Before patch.
* Before again.
* Using patch file to fix diesel_ltree issue with diesel.toml
* Adding some yalc ignores
* Fixing RegistrationMode enums
* Adding woodpecker diesel schema check.
* Try adding openssl 1.
* Try using diesel-cli image 1
* Try using diesel-cli image 2
* Try using diesel-cli image 3
* Try using diesel-cli image 4
* Try using diesel-cli image 5
* Try using diesel-cli image 6
* Try using diesel-cli image 7
* Try using diesel-cli image 8
* Try using diesel-cli image 9
* Try using diesel-cli image 10
* Try using diesel-cli image 11
* Try using diesel-cli image 12
* Try using diesel-cli image 13
* Making the chat server an actor.
- Fixes#2778
- #2787
* Forgot to add handlers folder.
* Some cleanup.
* Forgot to remove a comment.
* Address PR comments.
* Using ToString for enum operations.
* Fix listing type default value
The listing type parameter is only meant for the frontpage, but is
also applied inside of communities. The result is that this call
returns nothing, because it defaults to ListingType::Local:
https://fedibb.ml/api/v3/post/list?community_id=3
It needs to be called like this to get any posts:
https://fedibb.ml/api/v3/post/list?community_id=3&type_=All
This is clearly not expected behaviour, when a community is
specified, the listing type should default to All.
* fix clippy
* update activitypub-federation crate to 0.4.0
* fixes
* apub compiles!
* everything compiling!
* almost done, federated follow failing
* some test fixes
* use release
* add code back in
* Combine prod and dev docker setups using build-arg
- Fixes#2603
* Dont use cache for release build.
* Adding 2FA / TOTP support.
- Fixes#2363
* Changed name to totp_2fa for clarity.
* Switch to sha256 for totp.
* Get rid of Safe Views, use serde_skip
- Also change the ViewToVec, to work with non-vector cases. Might be
necessary in preparation for #2763
- Fixes#2712
* Forgot one safe
---------
Co-authored-by: Nutomic <me@nutomic.com>
* Show deleted and removed posts for profile views. Fixes#2624
* Only showing non-deleted/removed posts for creator.
* Add a admin or mod check to views, to show deleted and removed posts.
- Also removed the pointless "blanking" functions
* Fix clippy
* Make hidden posts comment clearer.
* Fixing federation tests.
* Fixing fmt.
This is not necessary because the domain cant change, so we only
need to insert if no row exists for this domain.
Also fetch instance actor when parsing person, not only community
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
* Implement separate mod activities for feature, lock post
Also includes collection for featured posts. Later we also need
to do the same for Comment.distinguished
* some changes
---------
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
* Combine prod and dev docker setups using build-arg
- Fixes#2603
* Dont use cache for release build.
* Separate comment distinguish into its own action.
- Fixes#2708
---------
Co-authored-by: Nutomic <me@nutomic.com>
- Other platforms can support additional language tags. Treat those
as None instead of throwing error
- deserialize_skip_error was implemented incorrectly and failed on
array values (6d9682f4e6)
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
The RFC 2822 format standard requires the message ID to be enclosed within angle brackets. If the standard is not followed, SpamAssassin deducts points from the e-mail.
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
* Speeding up comment-ltree migration, fixing index creation. Fixes#2664
* Adding some logging lines, fixing for missing posts.
* Adding more postgres config
* Post creation from Mastodon (fixes#2590)
* better logic for page title
* add deserialize helper
Co-authored-by: Dessalines <dessalines@users.noreply.github.com>
Dessalines